Show commands on groups of devices using Unicon and .yml hostfiles

Hello

After I created a script in Netmiko that connected to a number of devices to create an uptime report, I thought it would be nice to do a similar thing using Unicon. It was a much easier job.

here’s the python code that does the magic:

from unicon import Connection
from jinja2 import Environment, FileSystemLoader
import yaml
from pyats.topology import loader
import sys

def executeshowongroup(command,testbedfile,devicelist):
  realdevicelist = devicelist.split('%')
  print(realdevicelist)
  testbed = loader.load(testbedfile)
  x = 0
  
  for x in range(x,len(realdevicelist)):
    devicename = realdevicelist[x]
    c = testbed.devices[devicename]
    c.connect()
    c.execute(command)

if name == "main":
  command = sys.argv[1]
  testbedfile = sys.argv[2]
  devicelist = sys.argv[3]

executeshowongroup(command,testbedfile,devicelist)

and here’s the testbed.yml:

devices:
  myproxy:
   os: linux
   type: linux
   credentials:
    default:
     username: cisco
     password: cisco
    connections:
     cli:
      protocol: ssh
      ip: 1.1.1.1
      port: 722
   router1:
    os: ios
    type: router
    connections:
     defaults:
     class: unicon.Unicon
     cli:
      command: connect router1
      proxy: myproxy
   router2:
    os: ios
    type: router
    connections:
     defaults:
     class: unicon.Unicon
     cli:
      command: connect router2
      proxy: myproxy

Now I run the script with:

python3 unicongroupscript.py ‚show ip int brief’ newtestbed.yml router1%router2

It’s important to notice what is NOT in this script: gone are the time.sleep() commands, so i don’t need to worry if my proxy is particularly slow on any given day. I get the output much, much faster than in the case of netmiko. And the whole logic is done in python, unlike in Ansible where I’m artificially limited to the yaml syntax.

This is really, really cool stuff. One thing that is missing is the built-in group functionality in the testbed file, so scripts need to be run on individual hosts (or like in my case: with groups of hosts where delimiter between each host is %). I’ll try to talk to Cisco about this.